Speeding up the GENGA N-body integrator on consumer-grade graphics cards

Open Access

Table 1

Completion time in days for the various cards we tested.

GPU	FP32 (d)	FP32T (d)	FP64 (d)	FP64T (d)
1080 Ti	60.6	26.6	121.5	87.4
2070	69.4	30.6	173.3	129.5
3070	68.2	25.9	201.4	109.9
3090	51.4	19.0	101.7	65.5
K20	193.2	97.7	220.2	131.1
MI250	78.7	22.0	80.7	24.1
P100	77.6	19.1	77.7	21.0
A100	27.5	8.5	28.8	9.3

Notes. The T suffix means the tuned simulations. The drastic decrease in running time between the tuned and non-tuned simulations is caused by the tuning speeding up the first 10% of the simulation.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.